
\begin{table}[ht] \centering 
  \caption{Net-to-gross generation ratio regressions} 
  \label{tbl:reg_gr} 
\scriptsize 
\begin{tabular}{@{\extracolsep{2.0ex}}lD{.}{.}{-3} D{.}{.}{-3} } 
\\[-1.8ex]\hline 
\hline \\[-1.8ex] 
\\[-1.8ex] & \multicolumn{1}{c}{(1)} & \multicolumn{1}{c}{(2)}\\ 
\hline \\[-1.8ex] 
 Constant & 0.921^{***} & 0.897^{***} \\ 
  & (3,703.166) & (809.680) \\ 
  & & \\ 
 Grandfathering & -0.006^{***} & 0.002^{***} \\ 
  & (-18.617) & (4.666) \\ 
  & & \\ 
\hline \\[-1.8ex] 
Age &   & X \\ 
Size &   & X \\ 
Observations & \multicolumn{1}{c}{69,309} & \multicolumn{1}{c}{69,309} \\ 
R$^{2}$ & \multicolumn{1}{c}{0.003} & \multicolumn{1}{c}{0.050} \\ 
Adjusted R$^{2}$ & \multicolumn{1}{c}{0.003} & \multicolumn{1}{c}{0.050} \\ 
\hline 
\hline \\[-1.8ex] 
\multicolumn{3}{p{8.0cm}}{\textit{Notes:} This table reports results from two weighted least squares regressions based on Equation (\ref{Eq:net-to-gross}).  The dependent variable is net-to-gross  generation ratio, while the weights are based on monthly durations.  Column (1) essentially represents the pure weighted average, while Column (2) presents one conditional on age and size. They rely on CEMS and EIA-923 data from 2008 to 2017.  Boiler-level clustered standard errors used, with *** p$<$0.01,  ** p$<$0.05, * p$<$0.10 and \textit{t}-statistics in parentheses.} \\ 
\end{tabular} 
\end{table} 
